Can check mode be trusted when a playbook mixes core and third-party Ansible modules?

I want to validate an Ansible playbook against a staging inventory before any real run, using ansible-playbook --check --diff so nothing is actually modified. The playbook mixes core modules (template, service, package) with a few third-party collection modules that manage an external API.

My understanding is that each module decides whether it honors check mode via its supports_check_mode declaration, and modules that do not support it are skipped rather than executed. That raises a concern: a skipped task means later tasks using when conditionals or registered variables from that task may fail or evaluate differently than they would in a real run, making the simulation misleading.

Assume Ansible core 2.15+ with collections installed from Ansible Galaxy.

1. Is there a reliable way to identify which modules in a playbook will be skipped under --check before running it?
2. How should tasks that register variables from check-mode-skipped modules be structured so downstream when conditions still evaluate predictably?
3. Does --check combined with --diff still report accurate pending changes when earlier tasks in the play were skipped?